Development of Hazard Based-Models for the Metrorail of the Washington Metropolitan Area Transit Authority

摘要

The present study developed duration models to predict the duration of delay of the Washington Metropolitan Area Transit Authority's Metrorail based on incident data from June 2015 to May 2016. Three fixed-parameter hazard-based models (log-logistic, log-normal, and Weibull) were developed and compared. Out of the three models, the log-logistic gave the best results based on the Akaike information criterion (AIC) value and the log-likelihood values. From the results, incidents during peak hours, weekends, and on some specific metro lines yielded shorter delays. The results would enable transit authorities to identify the significant causes of delay to improve travel time reliability for transit riders.